ABSTRACT
Understanding crucial factors that affect the binding affinity of protein-RNA complexes is vital for comprehending their recognition mechanisms. This study involved compiling experimentally measured binding affinity (ΔG) values of 217 protein-RNA complexes and extracting numerous structure-based features, considering RNA, protein, and interactions between protein and RNA. Our findings indicate the significance of RNA base-step parameters, interaction energies, number of atomic contacts in the complex, hydrogen bonds, and contact potentials in understanding the binding affinity. Further, we observed that these factors are influenced by the type of RNA strand and the function of the protein in a protein-RNA complex. Multiple regression equations were developed for different classes of complexes to perform the prediction of the binding affinity between the protein and RNA. We evaluated the models using the jack-knife test and achieved an overall correlation 0.77 between the experimental and predicted binding affinities with a mean absolute error of 1.02 kcal/mol. Furthermore, we introduced a web server, PRA-Pred, intended for the prediction of protein-RNA binding affinity, and it is freely accessible through https://web.iitm.ac.in/bioinfo2/prapred/. We propose that our approach could function as a potential resource for investigating protein-RNA recognitions and developing therapeutic strategies.